Recommendations
AI-generated energy insights organized by severity — Critiques, Alertes, Normal, and Insights.
The Recommandations page lives under Optimisations in the sidebar. It surfaces AI-generated insights about your energy data, organized by severity level. The system automatically detects patterns and anomalies and presents them as actionable recommendations.
Severity levels
Critiques (Critical)
Issues that require immediate attention. Equipment failures, significant data gaps, or consumption anomalies costing real money right now.
Alertes (Alerts)
Important findings to review soon. Sustained consumption increases, equipment running outside expected parameters, or emerging trends that could become critical.
Normal
Observations about your energy usage worth knowing but not requiring urgent action.
Insights
Informational analysis and patterns. Seasonal trends, correlations between categories, or efficiency benchmarks relative to similar sites.
Recommendation structure
Each recommendation card includes:
| Field | Description |
|---|---|
| Category tag | Equipment category or area (e.g., CVC, Eclairage, Froid) |
| Site name | Which site the recommendation applies to |
| Description | Plain-language summary of what was detected |
| Analysis | Supporting data and context explaining why this matters |
| Suggested action | What you can do about it |
Discuter avec Rene (Discuss with Rene)
Each recommendation has a Discuter avec Rene button. Clicking it opens the Rene AI assistant panel with the recommendation pre-loaded as context. Ask follow-up questions like:
- "How much is this costing us per month?"
- "Has this pattern been getting worse over time?"
- "What would happen if we fixed this?"
How recommendations are generated
The system continuously analyzes your energy data across all sensors and categories. It compares current patterns against historical baselines, weather data, and expected operating profiles. When it detects a deviation worth reporting, it creates a recommendation with the appropriate severity level.
Recommendations refresh periodically as new data comes in. A recommendation may be promoted to a higher severity if the underlying issue worsens, or removed if the issue resolves itself.